Munich Personal RePEc Archive
Does Corruption Pay in Indonesia? If So, Who are Benefited the Most?
Pradiptyo, Rimawan
Faculty of Economics and Business, Universitas Gadjah Mada, Indonesia
17 September 2012
Online at https://mpra.ub.uni-muenchen.de/41384/
MPRA Paper No. 41384, posted 17 Sep 2012 13:34 UTC
Does Corrupt ion Pay in Indonesia ? If So, Who are Benefited the Most ? 1
Rimawan Pradiptyo2
Department of Economics, Faculty of Economics and Business Universitas Gadjah Mada
Indonesia
Abstract
This paper aims to assess the discrepancies in sentencing corruptors by judges in Indonesia’s judicial system. The data are based on the Supreme Court’s decisions during the period of 2001-‐2009 which available in public domain in www.putusan.mahkamahagung.go.id. The data comprise of 549 cases, which involved 831 defendants. The defendants have been classified into five groups depending on their alleged scales of corruptions (i.e. petty, small, medium, large and grand scale of corruptions).
The explicit cost of corruption during the period of 2001-‐2009 was Rp73.1 trillion (about US
$7.86 billion). In this paper, total financial punishment was estimated as the summation of the value of fines, seizure of assets (monetary only), and the compensation order sentenced by judges. The total financial punishment sentenced by the supreme judges during the period of 2001-‐2009 was Rp5.33 trillion (about US$573.12 million), therefore Rp67.77 trillion (US$7.28 billion) gap between the explicit cost of corruption and total financial punishment sentenced shall be borne by the tax payers.
Logistic and Tobin’s logistic (TOBIT) regressions have been used to analyse both the likelihood and the intensity of sentencing offenders, respectively, with particular punishments (i.e. imprisonment, fines, compensation order, etc.). The results show that the probability and the intensity of sentencing across various types of punishment do not correspond to the scale of corruptions. Offenders who committed petty and small scales corruption tend to be punished more severely than their medium, large and grand corruptors.
Keywords: Corruption, Court Decisions, Probability of Sentencing, Intensity of Sentencing, Logistic Regression, Tobin’s Logistic (TOBIT) Regression.
JEL Classifications: D02, D04, K14, K42
1 I would like to express my gratitude to conference participants in Kolkata, India, Perth, Scotland, Cambridge, UK for constructive feedback. I am indebted to Harry Gemilang, Seri Damayanti, Sony Sasongko for excellent assistantship in collecting the data. All remaining errors are my responsibility.
2 Contacting email address: Rimawan@gadjahmada.edu, Rimawan@feb.ugm.ac.id
1. Introduction
According to the utilitarian approach, the decision of a potential offender to commit an offence or not depends on the expected costs and benefits of the conduct. The expected costs of conducting an offence has been modelled as the interaction between any costs incurred (financially and non-‐financially) by the potential offenders if they have would have failed in committing an offence and the probability of being caught. Similarly, the expected benefits of conducting an offence can be estimated as the probability of success in conducting an offence and any gains (tangible and intangible) arose from conducting the offence. Becker (1968) used decision theory to analysed offenders and potential offenders behaviour. Excellent literature surveys in this area have been conducted by various authors including Garoupa (1997), Eide (2000, 2004), Bowles (2000) and Polinsky and Shavell (2000, 2007).
Another group of economists who use game theoretical analysis tend to be more pessimistic about the effectiveness of punishment as a mean to deter offending (Tsebelis, 1989). This article triggered a long debate involving several authors, including Bianco/Ordershook/Tsebelis (1990), Weissing and Ostrom (1991), Hirshleifer and Rasmusen (1992), Tsebelis (1990, 1991, 1992, 1993) and Andreozzi (2004). Recently Pradiptyo (2007) refined the inspection game proposed by Tsebelis, and showed that actually there is not so much discrepancy in the solution between decision theory and game theoretical approaches.
Irrespective of whether the approach is using either decision theory or game theory, it is assumed that potential offenders are rational. Individuals are going to commit an offence if the expected benefits of the activity exceed the expected costs of offending.
Consequently, in order to deter individual from committing an offence, the authority may increase the expected costs of offending bourned by potential offenders.
Attempts to increase the expected costs of offending can be done in several ways. The criminal justice authority may endeavour either to increase the probability of conviction, or alternatively, they may increase the severity of punishment. Indeed both possible scenarios are costly. In order to achieve the optimum level of deterrence, however, the criminal justice authority has two possible scenarios either by setting low probability of detection with high intensity of punishment or by setting high probability
of detection with low intensity of punishment (Becker, 1968, Garoupa, 1997, Garoupa and Klerman, 2002, 2004, Polinsky and Shavell, 2000, 2001, 2007).
A similar approach may be used in tackling corruptions. Any potential corruptors are rational individuals and accordingly they would conduct costs-‐benefits analysis prior to involve in corruptions. As applicable to other type of offences, the intensity of corruptions can be divided into several groups for instance small, medium and large scales of corruptions. The classification of the groups depends on the intensity of misallocation of resources owing to corruptions. Ideally, given the probability of detection and conviction, corruptors who committed larger scale of corruptions should receive sentence with higher intensity of punishment. In the case for which the courts determined to use financial punishment, then ideally a substantially higher intensity of financial punishment should be sentenced to more serious corruptors.
It should be noted that the characteristics of corruptors tend to be different in comparison to offenders conventional crimes. Table 1 provides comparison of characteristics between conventional offenders and corruptors. It may not be surprising, therefore, that combating corruptions is more difficult than tackling conventional crimes.
Table 1: Characteristics of Conventional Offenders and Corruptors
Conventional Offenders Corruptors
•The majority come from low income and low education background (Einat, 2004)
• In many cases they offended due to fulfilling necessities
•They come from high income and high education backgrounds
• Offending behavior is age sensitive (Bowles and Pradiptyo, 2005)
•The offending behavior is not age sensitive
•In many cases offenders were victims of bullying or crimes (Bowles & Pradiptyo, 2005)
•The use of sophisticated techniques which may be difficult to prove it
• The detection rate tend to be high • The detection rate tend to be lower since offenders may use their influence and power to prevent investigation
This paper aims to assess Indonesia’s court decisions in combating corruptions across various scales of corruptions. The data used in this study are based on the Indonesia Supreme court decisions from year 2001-‐2009. The dataset consists of 549 cases,
involving 831 defendants. All cases have been published in the official website of the Supreme Court in the following URL: http://putusan.mahkamahagung.go.id. The gravity of corruption and various anti corruption programs in Indonesia is discussed in section 2. Section 3 discusses the judicial system in Indonesia. Logistic and Tobin’s logistic regressions are used to evaluate the Supreme Court’s decisions. The model and the results of the analysis are discussed in section 4 and 5, respectively.
2. Corruption and Anti Corruption Programs in Indonesia
The Corruption Perception Index (CPI) in 2011 by the Transparency International placed Indonesia as the 100th country out of 183 countries in the world. In 2011, the CPI for Indonesia was 3.0, a small increase from CPI in 2010 that was 2.8. In 1999 the CPI of Indonesia was just 1.9 (See Figure 1). Indeed, according to the CPI, there is an improvement of condition in Indonesia, with respect to the perception of the subjects who take part in as respondents for developing the CPI. The improvement may not, however, necessarily sufficient to show the improvement in Indonesia.
Figure 1: The Corruption Perception Index (CPI) of Indonesia 1999-‐2011
Source: Transparency International, 1999-‐2011.
Recently, a survey by Hong Kong-‐based Political & Economic Risk Consultancy Ltd in 2010 scored Indonesia 9.07 out of 10.00 and placed Indonesia as the most corrupt country in Asia-‐Pacific region. This result was higher in comparison to 2009, which was
1.9 1.9 1.9 1.9 1.9 2 2.2 2.4 2.3
2.6 2.8 2.8 3
0 0.5 1 1.5 2 2.5 3 3.5
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
CPI of Indonesia 1999-‐2011
7.69. It turns out that problem of corruption in Indonesia is more acute then other countries in the region such as Cambodia, the Philippines, India, Thailand and Vietnam.
Corruptions in Indonesia had been flourished since the end of President Soekarno’s regime. Under President Suharto’s regime, corruptions had become spread out to all level of bureaucracy. After the end of President Suharto’s regime, reformations have been conducted in various fields, including politic, economic and also law. Anti corruptions programmes have been launched by the GoI post Suharto’s era, ranging from:
1. Ratification of Law no 31/1999 or Anti Corruption Act, which then be amended in 2001 by Law no 20/2001;
2. Ratification of Law no 30/2002 which mandated the establishment of Corruption Eradication Committee (KPK) and the KPK has been fully operated since 2004;
3. Ratification of Law 15/2002 of Anti Money Laundering Act, which mandated the establishment of Indonesian Financial Transaction Reports and Analysis Centre (PPATK) and the institution, has been fully operated since 2005. The act, then, was amended by Law no 8/2010.
4. In 2003 the Ministry of Finance has pioneered bureaucratic reformation which have been followed by other government departments up until now.
The Anti Corruption bill has been ratified in year 1999 and was refined in year 2001.
Indonesia has a penal law which is based in the Dutch penal law in 1811. Corruption has been considered as an extra ordinary crime; therefore it requires a special law to tackle it which is different from the Indonesia penal law.
In 2002, the GoI also ratified anti money laundering act, which is separate from the anti corruption act. The act provided the basis to form PPATK and the body has been fully functioning since 2005. Different from KPK, the PPATK does not have the power to bring defendants to courts. Instead, the PPATK functions as an intelligent unit, which provide information to law enforcement agencies such as police, KPK and office of prosecutor. Recently, the GoI refined the act in 2010 which provide a stronger position of PPATK to share any information that they obtained to other law enforcement agencies.
In year 2002, the parliament ratified a bill which became the foundation for the Corruption Eradication Committee. The Committee is an independent body which the main task is to tackle large-‐scale corruptions (i.e. Rp 1 billion or more). The Committee has been financed by government budget, however they report to the parliament and they do not report to President. The Committee has been fully operational since 2004.
Since then, corruptions have been dealt by two groups of law enforcers. For large scale corruptions (Rp 1 billion or more) have been tackled by the Committee, whereas for medium and small scales corruptions (less then Rp 1 billion) have been tackled by Police and Public Prosecutors. It should be noted that the committee may have a more powerful authority than the police in investigating corruptions. Furthermore the committee has been equipped with more sophisticated instrument which enable them to intercept any type of communication between suspects and their counterparts.
3. Judicial System in Indonesia
Indonesia follows continental law system and its’ penal code is based on 1881 Dutch penal code. Although, the Dutch has amended its penal code in 1994, the Dutch the penal code 1881 still has been implementing in Indonesia until now. It should be noted that the judicial system in Indonesia does not recognise the use of juries, instead the decisions whether a defendant guilty or not depends on the decisions of board of judges.
Under Indonesia criminal justice system, all criminal cases should be trialled before District Courts. Each District Court is situated in a Kabupaten (district) and there are 497 districts in Indonesia. Judges’ decisions in a district court may be appealed either by defendants or prosecutors if they dissatisfied with the decisions. In the event that the defendant does the appeal, which occurs in most corruption cases, then the case is referred to the High Court, which situated in the capital of each province. In the case for which the defendant does not satisfy with judges’ decisions in the High Court, a further appeal can be made to the Supreme Court. On the contrary, if the prosecutor does not satisfy with judges’ decisions in the District court, the case may be appeald directly to the Supreme Court.
After the Supreme Court sentenced the case, there is still an opportunity for conducting further appeal called a judicial re-‐examination by the Supreme Court. The judicial re-‐
examination can only be pursued if there is new evidence, which has not been put
before trial previously. It should be noted that the cost of court in Indonesia is economical. The judicial system in Indonesia rules that the there are three possible values of the court costs, namely Rp2500 to Rp10,000 (US$0.29 – 1.16), irrespective of how long the trials have been conducted.
Figure 2: Appeal Process under Indonesia Judicial System
Figure 2 shows the complexity of judicial system in Indonesia, starting from the detection by Police to the judicial review in the Supreme Court. The data used in this study are based on the Supreme Court decisions, both with and without any judicial review. Similar to other types of crime, the underlying number of corruptions is unknown in Indonesia. As the only information obtained was the Supreme Court decisions, any attempt to estimate the detection rate of corruptions would be daunting.
There are strong tendencies that appeal have been made up until the Supreme Court for the corruption cases3. In the case for which all corruption cases in the District Court have been appealed up to the Supreme Court, then the unobserved heterogeneity number 1 and 2 can be ignored. Nevertheless, points 3-‐5 are more serious and unfortunately the information may not be available. In essence the number of cases sentenced by the Supreme Court might be a tip of an iceberg of the underlying corruption cases in Indonesia.
3 Many thanks to Eddy OS Hiarej who informed me regarding this tendency.
Figure 3: Comparison of Appeal Cases in Indonesia and Other Countries
The conviction rate may be estimated from the data set and it should be noted that information in the Supreme Court decisions are very rich. Each the Supreme Court decision contains all information on the previous stage courts decisions. Therefore it is possible to trace back all information regarding the trials, evidence and also decisions in three different courts (i.e. District Courts, High Courts and the Supreme Court). By using a strict assumption that all corruptions cases put before the District Courts were appealed until the Supreme Court, then the conviction rate start from the State Court may be estimated4. It should be noted that none of the defendant or offender appeared more than one cases, therefore the data may not be able to support reconviction analysis.
Any attempt to analyse the data set may face unobserved heterogeneity issues. The unobserved heterogeneity arose from the fact that the data obtained record only any cases put before trials up until the Supreme Court. At least there are two potential source of unobserved heterogeneity. Firstly, a case in the District Court may be appealed either to the High Courts (by offenders) or to the Supreme Court (by
4 An informal discussion with an Indonesia penal law expert, Dr. Eddy OS Hiarej, strengtened the assumption that almost all corruption cases have been appealed.
Other&Countries& Indonesia&
Supreme' Court'
High' Courts'
District' Courts'
Supreme' Court'
High'Courts'
District'Courts'
2'
prosecutors). Secondly, some cases in the Supreme Court have been undergone a judicial review. In order to minimise the unobserved heterogeneity, both variables should be incorporated in the regression models.
Since the information is based on the Supreme Court decision, the analysis suffers from unobserved heterogeneity which were affected by several factors below:
1. The number of cases terminated up until the High Courts 2. The number of cases terminated up until the District Courts
3. The number of cases referred from Police to Prosecutors but not being prosecuted
4. The number of cases reported to and detected by police but not being processed or referred to prosecutors
5. The number of unreported/undetected corruptions
Under Indonesia’s penal code, the intensity of punishment should be stated clearly for each type of offences in the Bill. There are various type of punishments in the Bill including imprisonment, parole, fines, subsidiary of fines, compensation order, subsidiary of compensation order, the seizure of evidence, the court costs and other sentences (see Appendix A). In this study we defined financial punishment as the summation of money levied through fines, compensation order and the amount of money seized as evidence. The courts may seized other types of assets which were suspected as the result from corruptions, such as cars, houses, apartments, etc, however these assets were not included in our calculation due to its complexity.
The values of court costs5 and other sentences were also neglegible. The values of the court costs were either Rp2500 (US$0,27), Rp5000 (US$ 0,54) or Rp10,000 (US$ 1.08).
These values, suprisingly, applicable for any type of offences. Other sentences were not applicable for most offenders and there were complexity in converting the order to monetary value.
5 The court costs were either Rp2500 (US$0,25), Rp5000 (US$ 0,5) or Rp10,000 (US$ 1). These values applicable for all types of offences.
4. Model
The optimum deterrence effect of sentencing is subjects of two factors, namely the probability of conviction and the intensity of punishment. Irrespective of whether the analysis is based on decision theory or game theory, the deterrence effect of conviction arose from the combination of the both factors (Becker, 1968, Garoupa, 1997, Shavel and Polinsky, 2000, 2007, Pradiptyo, 2007).
The probability of conviction and also the probabilities of receiving a particular type of punishment have been estimated using Logistic regression. Logistic regression is part of limited dependent variable analysis, whereby the values of the dependent variable are binary (e.g. 1 or 0, yes or no, male or female, etc) as a function of a stream of explanatory variables. The result obtained from Logistic regression provides information on the direction and the level of significant of each explanatory variables in affecting the likelihood even in the dependent variable. Thus far, the coefficients in the Logistic regression do not mean anything apart from providing information on the direction and the significant of the variables. The contribution of each explanatory variables to influenced the dependent variable will be obtained if we estimate the marginal effect of the Logistic regressions.
The intensity of each punishment would be estimated by the use of Tobit Logistic (TOBIT) regression. The TOBIT analysis has been used since the value of dependent variable is bounded below, namely the data cannot be negative. As the minimum value of any type of punishment is zero, the parameter estimate would be biased if we use least square method. In order to overcome the problem, the TOBIT regression, which is part of maximum likelihood method, has been used to estimate the impact of various criminogenic factors to the intensity of various punishment.
Attempt will be made to present both Logistic and TOBIT regressions in a table, therefore the information on the probability of conviction and the intensity of sentencing can be observed and analysed simultaneously.
𝐷_𝑆𝐶_𝐺𝑢𝑖𝑙𝑡𝑦!=𝑎+𝑏
!𝐺𝑒𝑛𝑑𝑒𝑟!+𝑏
!𝐿𝑛(𝐴𝑔𝑒)!+𝑏!𝐿𝑛(SocCost)! +𝑏
!𝐷_𝑆𝑂𝐸_𝐸𝑚𝑝!+𝑏!𝐷_𝑀𝑃!+𝑏!𝐷_𝑃𝑟𝑖𝑣𝑎𝑡𝑒!+ 𝑏!𝐷_𝐽𝑎𝑤𝑎!
+𝑏!𝐷_𝐺𝑟𝑒𝑎𝑡𝑒𝑟𝐽𝑎𝑘𝑎𝑡𝑎!+𝑏!𝐷_𝐺𝑟𝑎𝑛𝑑_𝐶𝑜𝑟𝑟!+𝑏!"𝐷_𝐿𝑎𝑟𝑔𝑒_𝐶𝑜𝑟𝑟! +𝑏
!!𝐷_𝑆𝑚𝑎𝑙𝑙_𝐶𝑜𝑟𝑟!+𝑏
!"𝐷_𝑃𝑢𝑛𝑦_𝐶𝑜𝑟𝑟!+𝑏!"𝐷𝐶_𝐺𝑢𝑖𝑙𝑡𝑦!+𝑏
!"𝐷_𝐻𝑖𝑔ℎ𝐶𝑜𝑢𝑟𝑡!
+𝑏!"𝐷_𝐽𝑢𝑑𝑖𝑐𝑖𝑎𝑙_𝑅𝑒𝑣!
Whereby
D_SC_Guiltyi = Dummy variable whether the Supreme Court found defendant guilty (1 = Yes, 0 = Otherwise)
Gender = Gender of defendant (1 = Male, 0 = Female) Ln(Age) = Natural logarithmic function of age of defendant
Ln(SocCost)i = Natural logarithmic function of Social costs of corruptions estimated by prosecutors in nominal price (limited to explicit costs)
D_SOE_Empi = Dummy variable whether a defendant worked as State-‐Owned Enterprise’s Employee (1 = Yes, 0 = otherwise)
D_MPi = Dummy variable whether the defendant were Member of the Parliament (1 = Yes, 0 = otherwise)
D_Privatei = Dummy variable whether a defendant worked in private sector (1 = Yes, 0 = Otherwise)
D_ Jawa = Dummy variable whether the corruption was committed in the Island of Jawa (1 = Yes, 0 = otherwise)
D_GreaterJakarta = Dummy variable whether the corruption was committed in Greater Jakarta (1 = Yes, 0 = otherwise)
D_Grand_Corr = Dummy variable whether the defendant commited grand scale of corruptions, i.e. Rp25 Billion or above (1 = Yes, 0 = Otherwise) D_Large_Corr = Dummy variable whether the defendant commited large scale of
corruptions, i.e. from Rp 1 Billion to up to but not including Rp25 Billion (1
= Yes, 0 = Otherwise)
D_Small_Corr = Dummy variable whether the defendant commited small scale of
corruptions, i.e. Rp10 million to up to but not including Rp100 million (1 = Yes, 0 = Otherwise)
D_Petty_Corr = Dummy variable whether the defendant commited a petty scale of corruptions, i.e. up to but not including Rp10 million (1 = Yes, 0 = Otherwise)
DC_Guiltyi = Dummy variable whether District Courts found finesnt guilty (1 = Yes, 0 = Otherwise)
D_HighCourt = Dummy variable whether the case was appealed to the Supreme Court after being sentenced by the HighCourt (1 = Yes, 0 = Otherwise)
D_Judicial_Rev = Dummy variable whether after the Supreme Court sentenced the defendant, the decisions were requested to be reviewed.
In the model above, the decisions made by District and High Courts serve as independent variables. The aims of using this variable is to investigate the consistency between the decisions made by the District and the High Courts in comparison to the decisions of the Supreme Court.
The occupations of defendants were classified into four groups, namely Civil Servant, State-‐Own Enterprise’s Employee, Senator and those who worked in private sector. In this model, civil servant has served as a reference to the other occupations.
Furthermore, the corruptions were also classified into five different scales, namely grand, large, medium, small and petty corruptions. In the model above, the medium scale of corruptions has served a reference. The merit of using medium scale as a reference is the ability of the model to observe any difference in the intensity of punishment between large and grand corruptions in one side with petty and small corruptions on the other side. This approach enable us to deduce whether the court tend to treat different class of offenders differently.
In this study, the scale of corruptions have been classified into five groups, namely:
1. Petty corruption (up to but not including Rp10 million or US$1,075),
2. Small corruption (from Rp10 million to up to but not including Rp 100 million or US$10,753),
3. Medium corruption (from Rp 100 million to up to but not including Rp 1 billion or US$107,527),
4. Large corruption (from Rp 1 billion to up to but not including Rp 25 billion or US$2,688,172) and
5. Grand corruption (Rp 25 billion or above)
As previously discussed, the appeal system to the Supreme Court in Indonesia is quite unique. Not all cases which were appealed to the Supreme Court have got through High Courts. In order to observed possible unobserved heterogeneity among different routes of appeal to the Supreme Court a dummy variable named D_HighCourt was included in the model. Similarly another dummy variable named D_Judicial_Rev has been employed in order to observed possible variation in the probability of conviction whether or not the judicial review has been conducted to the initial Supreme Court decisions.
Similar to the regression model to estimate the likelihood of conviction by the Supreme Court, a similar approach was used to estimate the likelihood of offenders being sentenced by various types of punishments. The Logistic regression model of the likelihood of sentencing various types of punishments are summaries i the following equation.
𝐷_𝑆𝐶_𝑃𝑢𝑛𝑖𝑠ℎ𝑚𝑒𝑛𝑡
!
!
=𝑎+𝑏
!𝐺𝑒𝑛𝑑𝑒𝑟!+𝑏
!𝐿𝑛(𝐴𝑔𝑒)!+𝑏!𝐿𝑛(SocCost)! +𝑏
!𝐷_𝑆𝑂𝐸_𝐸𝑚𝑝!+𝑏!𝐷_𝑀𝑃!+𝑏!𝐷_𝑃𝑟𝑖𝑣𝑎𝑡𝑒!+ 𝑏!𝐷_𝐽𝑎𝑤𝑎!
+𝑏!𝐷_𝐺𝑟𝑒𝑎𝑡𝑒𝑟𝐽𝑎𝑘𝑎𝑡𝑎!+𝑏!𝐷_𝐺𝑟𝑎𝑛𝑑_𝐶𝑜𝑟𝑟!+𝑏!"𝐷_𝐿𝑎𝑟𝑔𝑒_𝐶𝑜𝑟𝑟! +𝑏
!!𝐷_𝑆𝑚𝑎𝑙𝑙_𝐶𝑜𝑟𝑟!+𝑏
!"𝐷_𝑃𝑢𝑛𝑦_𝐶𝑜𝑟𝑟!+𝑏!"𝐷𝐶_𝐺𝑢𝑖𝑙𝑡𝑦!+𝑏
!"𝐷_𝐻𝑖𝑔ℎ𝐶𝑜𝑢𝑟𝑡!
+𝑏!"𝐷_𝐽𝑢𝑑𝑖𝑐𝑖𝑎𝑙_𝑅𝑒𝑣!
D_SC_Punishmentij = Dummy variable whether the Supreme Court sentenced defendant i with punishment j
DC_Punishmentij = Dummy variable whether the Supreme Court sentenced defendant i with punishment j
The regression model in this analysis is similar to the regression model in the previous analysis, however, the difference lies in the sample of offenders who can be included for these analyses. The types of punishment are relevant only to those who were found guilty by the Supreme Court. Given that the subgroup of defendants were found guilty, the further question is which factors affect the likelihood of offenders were sentenced with a certain type of punishment as oppose to other possible punishments.
In order to estimatevarious factor which attributable to the intensity of each type of punishments sentenced to offenders, Tobin’s Lnit (TOBIT) analysis has been conducted.
The reason of using TOBIT regression is due to the fact that the intensity of punishment is always be positive or it cannot be lower than zero.
𝑆𝐶_𝑃𝑢𝑛𝑖𝑠ℎ𝑚𝑒𝑛𝑡
!
!
=𝑎+𝑏
!𝐺𝑒𝑛𝑑𝑒𝑟!+𝑏
!𝐿𝑛(𝐴𝑔𝑒)!+𝑏!𝐿𝑛(SocCost)! +𝑏
!𝐷_𝑆𝑂𝐸_𝐸𝑚𝑝!+𝑏!𝐷_𝑀𝑃!+𝑏!𝐷_𝑃𝑟𝑖𝑣𝑎𝑡𝑒!+ 𝑏!𝐷_𝐽𝑎𝑤𝑎!
+𝑏!𝐷_𝐺𝑟𝑒𝑎𝑡𝑒𝑟𝐽𝑎𝑘𝑎𝑡𝑎!+𝑏!𝐷_𝐺𝑟𝑎𝑛𝑑_𝐶𝑜𝑟𝑟!+𝑏!"𝐷_𝐿𝑎𝑟𝑔𝑒_𝐶𝑜𝑟𝑟! +𝑏
!!𝐷_𝑆𝑚𝑎𝑙𝑙_𝐶𝑜𝑟𝑟!+𝑏
!"𝐷_𝑃𝑢𝑛𝑦_𝐶𝑜𝑟𝑟!+𝑏!"𝐷𝐶_𝐺𝑢𝑖𝑙𝑡𝑦!+𝑏
!"𝐷_𝐻𝑖𝑔ℎ𝐶𝑜𝑢𝑟𝑡!
+𝑏!"𝐷_𝐽𝑢𝑑𝑖𝑐𝑖𝑎𝑙_𝑅𝑒𝑣!
where:
SC_Punishment = the intensity of punishment j sentenced to defendant i.
5. Results
Information from the dataset shows that the majority of defendants were male (93.1%) and only small fraction were female (6.9%). None of defendants who committed Grand scale alleged corruptions was female, however there were 45 male defendants (5,5%
)who were prosecuted for Grand scale corruptions . The number of defendants who were prosecuted for large corruptions were 201, of which 190 defendants were male (94.5%).
Table 2: Distribution of Defendants According to the Scale of Alleged Corruptions
Scale of Corruptions
Total Petty Small Medium Large Grand
Gender
Male 36 183 313 191 45 768
Female 2 16 29 10 0 57
Total 38 199 342 201 45 825
Location
Jawa 11 73 118 95 33 330
Greater Jakarta 0 5 18 53 27 103
Outside Jawa 27 124 224 105 12 492
Total 38 197 342 200 45 822
Occupation
Civil Servant 26 137 126 61 8 358
SOE Employees 1 9 33 25 12 80
MP 1 25 115 76 4 221
Private Sector 10 26 66 38 20 160
Total 38 197 340 200 44 819
Source: Indonesia Supreme Court, calculated.
Table 2 shows that more than 50% of defendants committed their alleged corruptions in outside Jawa. Of 330 alleged corruption cases in Jawa 31,2% have been committed in Greater Jakarta (GreaterJakarta6). There is a tendency that the grand-‐scale of corruptions were committed in Jawa, especially in Jakarta. This may not be surprising as Jakarta is the capital city and the centre of administration in Indonesia. About 90% of money has been circulated in Jawa and more than 47% of money has been circulated in Jakarta.
Civil servants tend to dominate petty and small scales corruptions as opposed to individuals from the other occupations. On the other hand, the defendants who worked in private sector dominate the alleged grand scale of corruptions. Indeed, the coverage of the anti corruption act in Indonesia is limited to civil servants, member of parliaments and also state-‐owned enterprise employees, howerver, individuals who work in private sector may become defendants as they may involve in corruption of government procurements.
6 This is stand for Jakarta, Bogor, Depok, Tangerang and Bekasi which comprises of 9 municipalities, which are Central Jakarta, South Jakarta, North Jakarta, West Jakarta, East Jakarta, Bogor, Depok, Tangerang and Bekasi.
Corruptions create misallocation of resources, therefore any attempt to estimate the cost of corruptions should be taken into consideration both explicit and implicit costs of corruptions. Unfortunately this is not the case in Indonesia as prosecutors, who mostly never received training in economics, have calculated the cost of corruptions limited to the explicit cost only. The consequences are that the costs of corruptions have been underestimated and there might be many cases of error types I and II in convicting defendants.
Table 3 shows that comparison between the total explicit costs, the total financial punishment prosecuted and total financial punishment sentenced by the Supreme Courts across various scales of corruptions. Offenders who commit petty scale of corruptions tend to be sentenced most severely than their counterparts. Although the total costs of corruptions they inflicted to society was Rp 93.4 million, they were prosecuted and sentenced for Rp1.7 billion (1800.3%) and Rp 1.2 billion (1234.8%), respectively. A similar anomaly occurs to offenders with small scale corruptions. The total financial punishment sentenced to them was more than double than that of prosecuted. The B:A ratio to this type of offenders was 186.6%, however the C:A ratio was 375.8%. Both types of offenders tend to be unfortunate as they received financial punishment more than the cost they inflicted.
The features of financial punishment sentenced for both petty and small scale corruptors may not be found on the other classes of corruptors. Indeed the medium scale corruptors were prosecuted for financial punishment for 120.9% above the cost of corruptions they inflicted. Nevertheless, the Supreme Court sentenced them with financial punishment worths 86.3% of their total cost of corruptions. The cost of corruptions attributable by this group was Rp84.8 billion, however they were sentenced with financial punishment worths Rp73.2 billion.
Table 3: Comparison between Cost of Corruption and Financial Punishment Sentenced
Scale of
Corruptions Offenders
Current Price
B:A (%) C:A (%) Explicit Cost of
Corruptions (A)
Total Financial Punishment Prosecuted (B)
Total Financial Punishment Sentenced by the Supreme Court (C)
Petty 22
Rp93,4 Million Rp1,7 Billion Rp1,2 Billion 1820.13% 1284.80%
($10,043.01) ($182,795.70) ($129,032.26)
Small 128
Rp5,1 Billion Rp9,6 Billion Rp19,3 Billion 188.24% 201.04%
($548,387.10) ($1.03 million) ($2.08 million)
Medium 240
Rp84,8 Billion Rp102,5 Billion Rp73,2 Billion 120.87% 86.32%
($9.12 million) ($11.02 million) ($7.87 million)
Large 122
Rp621,9 Billion Rp404,7 Billion Rp299,1 Billion 65.07% 48.09%
($66.87 million) ($43.52 million) ($32.16 million)
Grand 30
Rp58,09 Trillion Rp23,04 Trillion Rp3,95 Trillion 39.66% 6.80%
($6.24 billion) ($2.48 billion) ($424.73 million)
Total 542
Rp58,81 Trillion Rp23,55 Trillion Rp4,34 Trillion 40.04% 7.38%
($6.32 billion) ($2.53 billion) ($466.67 million) Scale of
Corruption Offenders Constant Price 2009 B:A (%) C:A (%)
Petty 22
Rp108,4 Million Rp1,8 Billion Rp1,2 Billion 1660.52% 1107.01%
($11,655.91) ($193,548.39) ($129,032.26)
Small 128
Rp6,3 Billion Rp11,6 Billion Rp25,4 Billion 184.13% 403.17%
($677,419.36) ($1.25 million) ($2.73 million)
Medium 240
Rp101,3 Billion Rp120,1 Billion Rp90,0 Billion 118.56% 88.85%
($10.89 million) ($12.91 million) ($9.68 million)
Large 122
Rp735,5 Billion Rp482,5 Billion Rp363,1 Billion 65.60% 49.37%
($79.09 million) ($51.88 million) ($39.04 million)
Grand 30
Rp72,22 Trillion Rp31,79 Trillion Rp4,87 Trillion 44.02% 6.74%
($7.77 billion) ($3.42 billion) ($523.66 million)
Total 542
Rp73,07 Trillion Rp32,41 Trillion Rp5,35 Trillion 44.35% 7.32%
($7.86 billion) ($3.48 billion) ($575.27 million) Source: Indonesia Supreme Court, estimated
Offenders who committed large and grand scales of corruptions tend to be more
‘fortunate’ than their counterparts who committed petty to medium scales of corruptions. The offenders who committed large and grand scales of corruptions were prosecuted with financial punishment about 65.07% and 39.66%, respectively, of their cost they have been inflicted to society. The ratio between the total financial punishment sentenced and the cost of corruptions decreased to 49.37% and 6.74%, respectively, for large and grand scale of corruptors, when they were sentenced by the Supreme Court. Imagine, 30 grand scale corruptors inflicted the cost of corruptions to
society worth Rp58.09 trilion, however the Supreme Court punished them with financial punishment worth Rp3.95 trillion (6.8%). If the estimation has been done in real price, then using price in 2009 as the constant price, then all offenders inflicted the cost of corruptions Rp73.07 trillion. Surprisingly, they were sentenced by the Supreme Court to pay the total financial punishment woths only Rp4.87 trillion (6.7%).
Table 4: The Comparison of Average Imprisonment Prosecuted and Sentenced
Types of Corruptions
Number of Offenders
Average Period of Imprisonment Prosecuted (month) [A]
Number of Offenders
Average Period of Imprisonment Sentenced
(mmonth) [B] B:A (%)
Petty 21 22.3 22 13.7 61.43%
Small 128 21.6 127 15.2 70.37%
Medium 237 53.2 240 32.8 61.65%
Large 122 79.0 122 43.5 55.06%
Grand 30 115.7 30 58.0 50.13%
Total 538 53.8 541 31.7 58.92%
Source: Indonesia Supreme Court, estimated
Further exploration on the sentencing for imprisonment found a similar pattern. Table 4 shows that, again, petty to medium scales of corruptors tend to be sentenced more severly in comparison to the other counterparts. The ratio of the average imprisonment sentenced to the average of imprisonment prosecuted by the Supreme Court were 55.0% and 50.1%, respectively, for both large and grand scales corruptors. In contrast, the same ratios were 61.4%, 70.3% and 61.6%, respectively for petty, small and medium scales of corruptors.
It should be noted that the length of imprisonment above was based on the Supreme Court’s decision and it did not reflect the actual length of imprisonment. The actual length of imprisonment tend to be shorter as every year, especially on the independence day, the government grants remission to offenders including corruptors. In general the actual length of imprisonment was about 60% of the Supreme Court’s sentencing.
The findings above give rise various unanswered questions which should be investigated further in the near future. Why do prosecutors and judges tend to treat offenders differently? Why do both petty and small scale corruptors tend to be treated harstly than the other counterparts? Why do prosecutors and judges tend to be much
more lenienced toward large and Grand scales of corruptors? What are the consequences which may arise due to the unfair sentencing as it was found above?
Table 5 provides information on various factors attributable to the probability of conviction in corruption cases in Indonesia. The result shows that the Supreme Court is highly likely to support District Courts’ decisions. A defendant who was found guilty by the District Courts is highly likely to be found guilty by the Supreme Court. Obviously any attempt to appeal is costly, however defendants tend to pursue to appeal when Distric Court decided that they were guilty.
Table 5: Logistic Regression of Conviction by the Supreme Court Logistic Regression
Dependent Variable: SC_GUILTY Included observations: 811
Variable Coefficient
Std.
Error Prob.
C 1.852 2.796 0.508
Gender 0.077 0.370 0.835
LN(Age) -‐0.810 0.506 0.110
LN(SocCost) 0.010 0.100 0.922
D_Jawa* 0.389 0.218 0.074
D_GreaterJakarta -‐0.076 0.383 0.843
D_SOE Empl*** 1.611 0.421 0.000
D_MP* -‐0.393 0.237 0.096
D_Private 0.334 0.264 0.206
D_Grand -‐0.258 0.795 0.745
D_Large -‐0.302 0.332 0.362
D_Small 0.032 0.322 0.921
D_Petty -‐0.347 0.621 0.576
D_Guilty_DC*** 3.236 1.136 0.004
D_Appeal_HC -‐0.622 1.137 0.584
D_JudicialReview*** 1.627 0.406 0.000 Source: Indonesia Supreme Court, estimated Note:
*) significant at α = 10%;
**) significant at α = 5%;
***) significant at α = 1%.
